Follow-up from "New flagged value masking behaviour"

The following discussion from !45 (closed) should be addressed:

• @schaefed started a discussion: (+3 comments)

  I am not entirely convinced, that moving the crucial part of evaluation into the module funcs makes it easier or more obvious to find. But the idea to make flags-to-mask a user settable options sort of requires, that the masking is part of the function wrapping.

  So, long story short, I think, that register.py should eventually move into the core, as it is going to become an integral part of the evaluation machinery.

  On more thing: getMask in snippet $49 is essentially a flagger.isFlagged on steroids, so I really think, we should implement the option to pass a (possibly empty) sequence of flags to flagger.isFlagged and reduce the snippet to:

  from lib.tools import toSequence
  
  def register():
      def outer(func):
          name = func.__name__
          func = Partial(func, func_name=name)
          FUNC_MAP[name] = func
  
          def inner(data, field, flagger, *args, **kwargs):
              # NOTE: would be nice to make to_mask a 'real' argument
              to_mask = toSequence(kwargs.get('to_mask', flagger.BAD))
              
              data_in = maskData(data, flagger, to_mask)
              data_res, flagger_res = func(data_in, field, flagger, *args, **kwargs)
              data_out = unmaskData(data, flagger, data_res, flagger_res, to_mask)
  
              return data_out, flagger_res
  
          return inner
  
      return outer
  
  
  def maskData(data, flagger, to_mask):
      data_masked = data.copy()
      data_masked[flagger.isFlagged(flag=to_mask)] = np.nan
      return data_masked
  
  
  def unmaskData(data_old, flagger_old, data_new, flagger_new, to_mask):
      mask_old = flagger_old.isFlagged(flag=to_mask)
      mask_new = flagger_new.isFlagged(flag=to_mask)
      mask_combined = DictOfSeries()
      for col, left in data_new.indexes.iteritems():
          right = mask_old[col].index
          # NOTE: ignore columns with changed indices (assumption: harmonization)
          if left.equals(right):
              # NOTE: don't overwrite flags that where unset within the called function
              mask_combined[col] = mask_old[col] & mask_new[col]
      data_new.aloc[mask_combined] = data_old[mask_combined]
      return data_new